BACKGROUND: There is no consensus regarding the definition of pediatric metabolic syndrome (MetS). This study assessed the impact of alternative definitions on the prevalence, children identified, and association with socioeconomic status (SES).
METHODS: Data were from the prospective multigenerational Dutch Lifelines Cohort Study. At baseline, 9754 children participated, and 5085 (52.1%) with average follow-up of 3.0 (SD = 0.75) years were included in the longitudinal analyses; median ages were 12 (IQR = 10-14) and 14 years (IQR = 12-15), respectively. We computed MetS prevalence according to five published definitions and measured the observed proportion of positive agreement. We used logistic regression to assess the SES-MetS association, adjusted for age and sex. Longitudinal models were also adjusted for baseline MetS.
RESULTS: MetS prevalence and positive agreement varied between definitions, from 0.7 to 3.0% and from 0.34 (95% CI: 0.28; 0.41) to 0.66 (95% CI: 0.58; 0.75) at baseline, respectively. We consistently found a socioeconomic gradient; in the longitudinal analyses, each additional year of parental education reduced the odds of having MetS by 8% (95% CI: 1%; 14%) to 19% (95% CI: 7%; 30%).
CONCLUSIONS: Alternative MetS definitions had differing prevalence estimates and agreed on 50% of the average number of cases. Additionally, regardless of the definition, low SES was a risk factor for MetS.
IMPACT: Little is known about the impact of using different definitions of pediatric metabolic syndrome on study results. Our study showed that the choice of pediatric metabolic syndrome definition produces very different prevalence estimates. We also showed that the choice of definition influences the socioeconomic gradient. However, low socioeconomic status was consistently a risk factor for having pediatric metabolic syndrome. In conclusion, studies using different definitions of metabolic syndrome could be reasonably compared when investigating the association with socioeconomic status but not always validly when comparing prevalence studies.